Community team Ingot's Qwen 27B FP8 quantization reportedly beats Alibaba's own
markjeffrey · x · 2026-10-22
Quantization community milestone: team Ingot (97) released Ingot-I, an FP8 quantization of Qwen 27B that reportedly outperformed Alibaba's own quantization, built in just one week. Next targets are 2-bit quantization and larger GLM models — a sign community quantization is now beating vendor defaults on some metrics.
More from Models
- Wikipedia goes down over two PHP flaws as OpenAI and humans spar over who found them — bindureddy · 2026-10-22
- Unsloth ships day-0 Dynamic GGUFs; local model nails tool calls, builds Flappy Bird in one shot — danielhanchen · 2026-10-22
- AISI backs Anthropic's Mythos: only model to finish a 32-step corporate network attack sim — dbreunig · 2026-10-22
- Reflection ships open Beam customization recipes spanning LoRA, SFT and GRPO RL — HowDevelop · 2026-10-22
- KRAFTON drops gate for Raon-1 mini, a 30B-A4B open MoE model under Apache 2.0 — Kangwook_Lee · 2026-10-22
- Devs push back on Grok's X data moat hype: browser use and API already cover it — brandon_galang · 2026-10-22